English liquid dissimilation

[This is an expanded version of a query I posted to Linguist List a few days ago.]

I’m interested in what appears to be a sporadic but widespread pattern of liquid dissimilation in English. I’ll post here the data I’ve collected so far, in hopes that readers might think of more examples or just want to share their wisdom on the subject.

Many English words can be heard at least occasionally with what appears to be a dissimilatory deletion of /r/. Some examples that have been reported include the following (many of these come from a website by an amateur linguist. Others are my own observation or have been contributed by other linguists.)


advesary, apeture, barbituate, beserk, bombadier, camaderie, celebatory, coner, cooborate, deteriate, entepreneur, extordinary, fustrated, govenor, hiarchy, impopriety, infastructure, interpet, itinery, Labador, libary, litature, onery, paraphenalia, paticular, peripheal, pespiration, perogative, proprietess, prostate, purient, quater, repetoire, respitory, secetary, spectogram, suprise, tempature, terrestial, terrist, themometer, Tristam Shandy, tumeric, venacular, vetinarian

There are a few cases where /r/ turns into something else: Feb[j]uary, defib[j]ulator for defibrillator, and I’ve heard reports of f[l]ustrated.

I’ve also noticed one example of dissimilatory /l/ deletion. I always thought Pachelbel’s Canon was by Pachebel. A google search shows I’m not the only one. I even get a morphological /r/-zero alternation with govern / government / governor. The /r/ is present for me in the first two, but optional in the last.

The phenomenon is mentioned by H.L. Mencken in The American Language (in the section Consonants). He cites an earlier work which I haven’t gotten ahold of yet: Hempl, George (1893) “Loss of r in English through dissimilation”. Dialect Notes, Vol. 1, pt. 6, pp. 279-81.

I’m leaning towards thinking that many of these are best explained by an Ohala-style analysis of dissimilation through misperception. As Mark Jones pointed out to me, a lot of the examples involve schwa-/r/ or /r/-schwa sequences, which in rapid speech would be pronounced as a rhoticized schwa. Listeners may assume the rhoticity comes from another source, namely the other /r/ in the word, and reanalyze the rhoticized schwa as a plain schwa.

7 thoughts on “English liquid dissimilation

  1. Bridget Samuels

    I can add a couple: inf(r)ared and p(r)oportional. It seems like the elimination of -rVr- sequences can be analyzed as what Blevins would call “chance” reinterpretation. This might not be so easy to posit for the long-distance cases, though.

  2. AJD

    A class project that’s regularly assigned for Penn’s intro-sociolinguistics class for undergrads is to find out to what extent non-rhoticity exists in Philadelphia using a couple of Philly street names. R-deletion is found in Girard with about twice the frequency of Market.

    Also: odinary.

  3. Nancy Hall

    Thanks! Those are great examples. I especially like infared; I can totally say that.

    The Girard / Market difference raises an interesting question. According to Ohala’s theory, the reason for dissimilation is that listeners are confused by sounds that have long-distance spectral effects. In the case of /r/, rhoticization spreads across much of the word, and the listener is unsure whether there is one source of rhoticity or two. In a word with two /r/s, the listener might wrongly attribute the spectral influences of one /r/ to the other, and hence perceptually ‘delete’ one of the /r/s.

    So when undergrads report hearing /r/ deletion in Girard, is it because the second /r/ is really not there, or because the undergrads can’t distinguish it? In other words, are the speakers using /r/ deletion as an alternative way to avoid multiple /r/s, or are the listeners doing the kind of misperception that Ohala predicted?

  4. Michael Covarrubias

    The one person I knew who said “flustrated” used the word in a context that led me to think he was combining flustered and frustrated. I don’t think he did it knowingly. He also said “roley-coaster” which sounds to me like vigilant dissimilation. Maybe the schwa wasn’t different enough so the [i] became necessary.

  5. Nancy Hall

    Thanks, Michael. I’d never heard “rollycoaster”, but it appears in the CoalSpeak Dictionary, which claims to describe speech in the “Coal Region” of Pennsylvania. I’d guess the [i] could also be partly a morphological reanalysis, on analogy with words like roly-poly.

  6. Pingback: phonoloblog»Blog Archive » celebu-blends

Leave a Reply

Your email address will not be published.